Delta DPS-160GB

[linuxguru]

I've opened up a rare failed HP (Delta OEM) mini-ATX PSU, model DPS-160GB, HP P/N 0950-3751, 185W. It seems to be a PII/PIII-era PSU for a small tower case like a Brio or maybe a Vectra. All caps are Nippon Chemicon, secondaries are mostly LXJ, no visible failures. The primary-side circuit topology seems to be a single-ended forward converter (with a single Toshiba n-channel 2SK2611 power-FET as the main switch). There's a daughter-card on the secondary side with the controller electronics (including some possibly custom ICs).

Upon plugging it in, +5Vsb is present and good. However, it fails to power up either when connected to a mobo (by momentarily shorting the power-on switch) or by shorting PSON to ground. There's no visible sign of any damage or fire/smoke on any component. The main power FET has not been checked out yet, because it's tricky to remove it. Secondary rectifiers haven't been checked yet, either.

What are most common failure modes I need to check for first?

[Rainbow]

Re: Delta DPS-160GB

I first check if +5V or +12V isn't shorted to ground - that means shorted rectifier.

[PeteS in CA]

Re: Delta DPS-160GB

If you ground the On/Off pin, you can troubleshoot it without being connected to the MB. It should operate normally. The fact that you have +5V Stby means that the main fuse is not blown, so be careful around the primary side, where you have line potential AC voltages and ~300VDC. If you have an oscilloscope, check the +3.3V, +5V and +12V when you enable the P/S. If you see any of them overshoot severely (>4V, >6V or >13.5V), then that O/P that overshoots is tripping over-voltage protection, and the problem is in that regulator circuit. OTOH, if it does nothing at all, there is a resistor (~47K-150K) and a small electrolytic capacitor (~22uF-100uF) that provide start-up bias to the PWM IC (controller, probably a UC384X family part). If either are open, the P/S main regulator won't start up. The +5V Stby is a totally separate regulator and transformer, and thus can be operational while the main regulator does not. Unless HP has used this P/S for several generations of computers, I would guess that this P/S was built in the 1999-2001 timeframe.

[Shroomie]

Re: Delta DPS-160GB

I've had two of these. Mine were from Pavilion 8750c (P3-667) and the next model up (Athlon Pluto 850, 8770c I think). 3.3v is 16a, 5v is 25a, and 12v is 4a, right?

Unfortunately, both of mine still worked, so I can't help you there...

[linuxguru]

Re: Delta DPS-160GB

> I first check if +5V or +12V isn't shorted to ground - that means shorted rectifier.

Right on the money - it looks like the resistance from +5v to ground is too low, though not an outright short. One of the pair of diodes on the rectifier must have bitten the dust - it's *really* painful to replace. The heatsink, as well as all the 4 devices mounted on it have to be de-soldered. There's also a fair amount of white epoxy to be cleaned up.

> there is a resistor (~47K-150K) and a small electrolytic capacitor (~22uF-100uF) that provide start-up bias to the PWM IC (controller, probably a UC384X family part).

The topology is slightly unusual, in that the PWM controller is entirely on the secondary side, even though this is only a single-ended forward (for which the controller is traditionally left on the primary side). I'll get around to checking the PWM controller card after clearing the rectifier issues.

The build quality is excellent, as is usual with Delta. The rectifier failure must have been due to an overload and/or mobo short, or maybe a fan failure.

> 3.3v is 16a, 5v is 25a, and 12v is 4a, right?

Exactly right.

[Shroomie]

Re: Delta DPS-160GB

To add, both motherboards attached to these Deltas had Rubycon capacitors (forget the models, I'll check if you want). The AMD was a modified ASUS K7M, the other one I forget. Both of them were going strong, the AMD board is still semi-in-use, now hooked up to a 235w Power Man (Fortron built).

If you want, I can supply photos with mediocre lighting at up to 2 megapixels.

[linuxguru]

Re: Delta DPS-160GB

Just checked the rectifier on the +5v rail by disconnecting the common cathode (took about 15 mins and lots of desoldering braid to mop up the solder completely around the lead) - it tests OK, open when reverse-biased and a very low forward resistance (typical of Schottky rectifiers). I'll repeat this test on all the rectifiers first.

> OTOH, if it does nothing at all, there is a resistor (~47K-150K) and a small electrolytic capacitor (~22uF-100uF) that provide start-up bias to the PWM IC (controller, probably a UC384X family part).

OK, found this. The primary-side controller is a UC3843 on a very small daughtercard - I missed it the first time around. I'll check this is well. BTW, I don't suppose there's any shut-down interlock that prevents startup with the fan disconnected, is there? I had to disconnect the fan to have easy access to the board.


> If you want, I can supply photos with mediocre lighting at up to 2 megapixels.

Thanks, no need - there's no visible problem to compare.

[larrymoencurly]

Re: Delta DPS-160GB

If that white stuff isn't shiney, it's not epoxy but something that can be dissolved with lacquer thinner, carburetor spray, etc. Unfortunately those chemicals also dissolve a lot of plastics, like vinyl.

One of my Deltas won't start without a load on the +5.0V, but another model also needs a load on the +3.3V to run.

[linuxguru]

Re: Delta DPS-160GB

> a small electrolytic capacitor (~22uF-100uF) that provide start-up bias to the PWM IC (controller, probably a UC384X family part).

The capacitor is a 47uF, 35V cap that's charged from a flyback winding from the Auxiliary/Standby transformer. I replaced it by way of abundant precaution with a 100uF, 35V. Upon powering up the aux supply, I measure +10.5V on it (and +4.99V on +5Vsb), which goes to the daughtercard with the UC3843. That seems to be above the UVLO threshold (7.8 to 9V) for the UC3843 to operate (though the nominal Vcc is specified as +15V in the datasheets). Does 10.5V seem like a reasonable operating voltage for a UC3843 driving a 2SK2611?

> One of my Deltas won't start without a load on the +5.0V, but another model also needs a load on the +3.3V to run.

Ah, this may be an issue. I probably need to hook it to a mobo for it to start.

[kc8adu]

Re: Delta DPS-160GB

thats a bit low.it may work but most i have seen with uc38xx ran 14-17v.
if the unit needs a load it will start and die.
i would replace the 3843 as its likely bad.
loading its supply down.